Skip to content

Multilingual embeddings: setup guide + cross-lingual eval #2

Description

@ahmetvural79

Cross-lingual retrieval (ask in one language, hit a document in another) comes from the embedder, not keyword search. docs/PROVIDERS.md covers the what but not an end-to-end setup + an eval that proves it.

Scope

  • step-by-step OpenAI and Ollama embedder setup
  • a small mixed-language fixture brain
  • an eval asserting a query in language A retrieves a source in language B
  • note PGLite has no stemmer/unaccent (vector signal carries morphology)

Metadata

Metadata

Assignees

No one assigned

    Labels

    documentationImprovements or additions to documentationenhancementNew feature or request

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions